Key points

Introduction

Human papillomavirus (HPV) is a common sexually transmitted infection that is the cause of a distinct subset of oropharyngeal cancer (OPC) rising in incidence in the United States and other developed countries. This increased incidence, combined with the strong effect of tumor HPV status on survival, has had a profound effect on the head and neck cancer (HNC) discipline. The multidisciplinary field of HNC is in the midst of re-evaluating evidence-based algorithms for clinical decision making, developed from clinical trials conducted in an era when HPV-negative cancer predominated. This article reviews relationships between tumor HPV status and gender, cancer incidence trends, overall survival, treatment response, racial disparities, tumor staging, risk stratification, survival post disease progression, and clinical trial design. Elucidation of the causal role for HPV in HNC has already altered the understanding of risk factors for HNC, and we anticipate a time in the near future when it will have a tremendous impact on clinical decision making.

Introduction

Human papillomavirus (HPV) is a common sexually transmitted infection that is the cause of a distinct subset of oropharyngeal cancer (OPC) rising in incidence in the United States and other developed countries. This increased incidence, combined with the strong effect of tumor HPV status on survival, has had a profound effect on the head and neck cancer (HNC) discipline. The multidisciplinary field of HNC is in the midst of re-evaluating evidence-based algorithms for clinical decision making, developed from clinical trials conducted in an era when HPV-negative cancer predominated. This article reviews relationships between tumor HPV status and gender, cancer incidence trends, overall survival, treatment response, racial disparities, tumor staging, risk stratification, survival post disease progression, and clinical trial design. Elucidation of the causal role for HPV in HNC has already altered the understanding of risk factors for HNC, and we anticipate a time in the near future when it will have a tremendous impact on clinical decision making.

Human papillomavirus biology

HPVs are small, nonenveloped, double-stranded DNA viruses of approximately 8000 base-pairs in size that have a distinct tropism for human epidermal and mucosal epithelium. More than 150 HPV types, distinguished by viral sequence variation, have been isolated from humans to date. Most infections spread by direct human-to-human contact are asymptomatic. Viruses with low oncogenic potential, called low-risk types, may cause benign hyperproliferation of the epithelium that manifests as skin warts, genital warts, or oral papillomas. The viral genome encodes early proteins that promote viral maintenance in the infected cell nuclei and viral replication (E1, E2, E4, E5, E6, and E7). Late proteins L1 and L2 encode the viral capsid proteins.

HPV types are classified as oncogenic or “high-risk” based on case-control studies demonstrating significant associations with cervical cancer. High-risk types include 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, and 66. However, HPV types 16 and 18 are accountable for approximately 70% of cervical and greater than 90% of noncervical cancers (eg, vaginal, vulvar, penile, anal) caused by HPV, including HNCs. The transforming ability of high-risk HPV types has been attributed to myriad functions of the classic viral oncoproteins E6 and E7, which disrupt regulation of cellular replication and differentiation to facilitate viral replication. HPV E7 oncoprotein disrupts control of the G1 to S phase of the cell cycle through interactions with pRb family members (p105, p107, p130), and HPV E6 prevents apoptosis by inducing degradation of p53. HPV16 E6 and E7 expression is sufficient to immortalize human keratinocytes, but insufficient for malignant progression. HPV E6 and E7 promote genomic instability, leading to secondary genetic events necessary for malignant progression. However, these secondary events remain poorly defined.

Early in infection, HPV genomes replicate as extrachromosomal elements in the nucleus. The frequency of viral integration increases with severity of precancerous lesions, and most cervical cancers harbor HPV integrants. The literature suggests approximately 20% to 48% of HPV-positive OPC have HPV integrants, but the sensitivity of assays used to detect HPV integrants in these studies has been called into question. HPV integration imparts genomic instability and a selective growth advantage on infected cells. Enhanced expression and stabilization of viral oncogene transcripts and disruption of the viral repressor HPV E2 contribute to such clonal selection.

Until recently, insertional mutagenesis has not been widely accepted as a functionally important consequence of HPV infection. However, Akagi and colleagues recently reported a striking association between HPV integrants and focal host alterations in genomic structures, frequently disrupting the expression and function of cancer-associated genes. Subsequently, HPV integration was associated with copy number variation in cervical cancer. Moreover, expression of genes neighboring HPV integration sites was significantly higher than that found in tumors without HPV integration. Recently, HPV integration analysis in cervical dysplasias, cancers, and cell lines identified recurrent genomic HPV integration hotspots. Analysis of 35 HPV-positive OPC in the TCGA dataset revealed an association between HPV integration and genomic structural variation and alterations in host cell gene transcription. The potential role of HPV integration and associated alterations of the host cell genome in cancer progression and the malignant phenotype is an active area of investigation.

Epidemiology of human papillomavirus–positive oropharyngeal cancer

In 2007, the International Agency for Research on Cancer stated that there was sufficient evidence to conclude that HPV16 is a cause of OPC and oral cavity cancers in humans. A review of the molecular and epidemiologic evidence in support of this causal association was updated by Gillison and colleagues in 2012 and 2015. In brief, strong and consistent associations between high-risk HPV infection and OPC have been observed in numerous case-control studies. In nested case-control studies, serologic evidence of HPV16 infection has been estimated to elevate risk of OPC by 14 to 270 times. The original observations reported by Gillison and colleagues of a causal association between predominantly HPV16 and a distinct molecular, clinical, and pathologic subset of OPC with an improved prognosis have now been reproduced on a global scale. Castellsague and colleagues collected 3741 HNCs diagnosed after 1990 from 32 countries and have reported preliminary analysis of tumor HPV status by HPV DNA, E6/E7 mRNA, and p16 immunohistochemistry. Most (∼88%) HPV-positive samples had type HPV16 DNA. HPV presence was associated with nonkeratinizing histopathology and location in the lingual and palatine tonsils of the oropharynx. This study estimated that approximately 18% of all OPCs worldwide were attributable to HPV infection (as defined by combined HPV DNA, E6/E7 mRNA, and p16 IHC positivity).

Case-control studies have estimated individuals with an oral HPV16 infection to have a 50- to 200-fold increased risk of OPC. The behavior most strongly associated with odds of OPC and prevalent oral HPV infection is oral-genital sex. The odds of cancer and infection increase with increased number of lifetime partners for oral-genital sex. Prospective studies of the natural history of oral HPV infection have now temporally linked oral-genital sex with risk of acquiring an oral HPV infection.

Recent studies of the epidemiology of oral HPV infection in the US population have observed the distribution of high-risk HPV infection to mirror incidence and risk for OPC in the United States. In 2009 to 2012, the prevalence of oral high-risk and HPV16 infections in the United States was three- to five-fold higher among men than women, consistent with a three- to five-fold relative risk for OPC among men versus women. Prevalence also increased with age, as does risk of OPC. Male gender, increasing age, intensity of current cigarette smoking, and number of lifetime oral-genital sexual partners were independently associated with prevalent high-risk oral HPV infection. Prevalence was therefore highest among men, aged 60 to 69 years, who were current smokers of greater than 20 cigarettes per day with a history of greater than 20 lifetime sexual partners. However, the number of individuals who fit this risk-profile in the US population was low. When the absolute number of individuals in the US population with an oral high-risk infection was considered, the burden of infections was greatest among men aged 40 to 59 years who were never or former smokers with a moderate number of sexual partners (2–19). This explains the frequent presentation of HPV-positive OPC in this subgroup of the US population.

Importantly, the study of oral high-risk HPV infection in the US population also provided a possible explanation for the escalating risk of OPC since 1984 in the United States (and other developed countries ) among men, but not women. Although men reported a higher median lifetime number of sexual partners than women, oral HPV prevalence differences by gender were largely attributable to stronger associations between sexual behavior and infection among men. For example, the increase in HPV prevalence per oral-genital sexual partner was three-times stronger in men versus women, and prevalence plateaued after 15 partners among men in comparison with only five partners among women. That is, women seemed to be protected from increased HPV exposure after five sexual partners, whereas men were not. Lower rates of seroconversion and development of protective neutralizing antibodies in response to genital HPV infection in men versus women are hypothesized to explain this observation. Thus, recent birth cohorts of men may have experienced a disproportionately greater impact than women of sexual behavioral changes associated with oral HPV exposure, resulting in disproportionate increased OPC incidence largely among men.

Human papillomavirus tumor status, race, and survival

In the 1970s, incidence rates for OPC were significantly higher among black persons than white persons in the United States. Over the four subsequent decades, rates significantly declined for black men but increased for white men, resulting in significantly higher rates among white men by 2013. An analysis of tumors collected through US Surveillance, Epidemiology, and End Results (SEER) registries from 1984 to 1999 revealed nonsignificant increases in HPV-prevalence of OPC among black persons in the United States, in contrast to strong increases among white persons. These data indicate that increased HPV prevalence in OPC over time among black persons reported by some institutions was likely attributable in large part to reduced incidence rates for HPV-negative cancers. By contrast, increased HPV prevalence in OPC among white persons was attributable to increased incidence for HPV-positive OPC. In agreement with these data, several single-institutional case series (but not all ) have observed higher HPV prevalence among OPC diagnosed among white than black persons. Indeed, an analysis of 557 OPC collected by US SEER cancer registries from 1995 to 2005 observed lower prevalence of high-risk HPV DNA among non-Hispanic blacks (∼50%) than all other races (>73%).

Historically, survival rates from HNC have been lower among black persons in comparison with white persons in the United States. These differences have been attributed to higher stage at presentation, comorbidities, and socioeconomic factors that affect access and delivery of care. Recently, several single-institutional analyses have investigated the impact of tumor HPV status on survival outcomes from HNC by race (black vs white). Generally, HPV-positive tumors were more frequent in white than black persons and were associated with a strong survival benefit. Most of these studies conclude that race was not an independent predictor of poor survival for OPC after accounting for tumor HPV status. For example, a shorter median survival among black than white persons with HNC enrolled in the TAX 324 was not observed after stratification by tumor HPV status. Similarly, among OPC patients treated in the Veterans Affairs system, race was not associated with survival after accounting for HPV status. It should be noted, however, that many of these studies either did not investigate (or were underpowered to do so) the independent effect of race after adjustment for known prognostic factors (including HPV) in multivariate models.

Several studies suggest that differences in tumor HPV status by race do not entirely explain survival differences, even in patients with OPC. In a prospective analysis of a uniformly treated patient population enrolled in Radiation Therapy Oncology Group (RTOG) 0129, black race remained associated with significantly lower survival after chemoradiotherapy for OPC patients and HNC patients overall, after adjustment for tumor HPV status, tumor and nodal stage, age, and treatment assignment. A simulated analysis of US SEER data observed that differences in survival by race observed for OPC patients may be attributable to differences in tumor HPV status by race, but that survival differences by race persisted for most non-OPC sites. Indeed, differences in tumor HPV status and smoking by race accounted for the differences in survival for OPC patients, but not for non-OPC cancers in another analysis. An analysis of a case series conducted at the University of Maryland also observed that HPV-negative tumor status and black race remained independently associated with less favorable survival, largely because of poorer survival among black persons with HPV-negative tumors. There is thus a growing consensus across studies that race remains a negative prognostic factor for survival for patients with HPV-negative cancers. Although differences in survival by race for patients with OPC are in large part attributable to differences in tumor HPV status by race (with HPV-positive status more common among white persons), racial disparities in outcome are still present for patients with OPC, but seem significantly larger in absolute magnitude for non-OPC.

The underlying reasons for the difference in HPV prevalence in OPC by race are in large part explained by higher incidence for HPV-negative cancer among black persons and HPV-positive OPC among white persons. In 1993, Day and colleagues reported that higher incidence rates for oral cancers among black than white persons was caused by a higher frequency of heavy drinking and current smoking among black persons and found no differences in estimates for relative risks for these exposures by race. Although current tobacco smoking declined from 1965 to 2007 in the United States for black and white persons, smoking rates remain higher for black persons. These declines are consistent with declining incidence rates for HPV-negative OPC in the United States. At this time, the relative increase of HPV-positive OPC among white versus black persons remains unexplained. Interestingly, increasing incidence trends from 1978 to 2007 in the United States for HPV-related preinvasive and invasive anogenital (eg, cervical, anal, vulvar, penile) cancers were similar among white and black persons. These data suggest changes in genital HPV infection trends were similar over time among black and white persons. Although ever performance of oral-genital sex and number of lifetime oral sex partners were higher for white than black persons in the United States, the prevalence of HPV16 or high-risk oral HPV infections in 2009 to 2010 was nonsignificantly higher in white persons. Frequent, concomitant age at first experience of oral and genital sexual behaviors among white persons, in contrast to sole genital or sequential behaviors in black persons, could in theory result in higher protective seroconversion rates after genital infection in black persons, resulting in relative protection from subsequent oral HPV infection.

Human papillomavirus status, tumor staging, and prognosis

An analysis by Ang and colleagues conducted within RTOG 0129 demonstrated that tumor HPV status and tobacco smoking history had a more significant impact on patient prognosis than American Joint Committee on Cancer (AJCC) TNM stage. These data raise the question as to whether or not the current staging system should be modified to reflect differences in survival for HPV-positive versus HPV-negative OPC by stage. Single institutional retrospective analyses at MD Anderson Cancer Center and Princess Margaret Hospital revealed the current AJCC TNM stage to inaccurately reflect survival rates for all patients with OPC. Survival rates declined in a stepwise manner as a function of advancing AJCC TNM stage for HPV-negative OPC patients, but not for HPV-positive OPC patients (using p16 IHC as a surrogate for HPV tumor status). Recursive partitioning analysis demonstrated that survival was superior for patients diagnosed with HPV-positive, low-volume (eg, T1-T3, N0-N2c) as compared with high-volume local-regional disease (eg, T4 or N3 ). Tobacco smoking and age further stratified survival: tobacco smoking history greater than 20 pack-years reduced survival for low-volume disease, whereas age greater than 70 years similarly affected high-volume disease.

Although there is growing consensus that the AJCC staging system will require modification to reflect the differences in prognosis by stage for HPV-positive and HPV-negative OPC, additional research is needed to determine specifics of these modifications. Notably, the different prognostic groups defined within retrospective analysis of clinical trials and single-institutional case series are being used to determine clinical trial eligibility for treatment deintensification protocols. For example, only patients with low-volume (eg, T1-T3, N0-N2b) HPV-positive OPC with low tobacco smoking exposure (<10 pack-years) are eligible for treatment deintensification protocols currently being conducted by NRG Oncology and Eastern Cooperative Oncology Group (ECOG)/American College of Radiology Imaging Network. Therefore, we can anticipate that additional data relevant to AJCC staging categories together with results from these clinical trials will eventually result in stratification by tumor HPV status of the AJCC staging system and National Comprehensive Cancer Center practice guidelines.

Human papillomavirus status, surgical resection, extracapsular extension, and risk of cancer progression

The presence of extracapsular extension (ECE) in surgically resected cervical lymph nodes is a well-established biomarker for risk of local-regional recurrence, distant metastases, and reduced survival. Randomized controlled trials (RTOG 9501 and European Organization for Research and Treatment of Cancer [EORTC] 22931) demonstrated significantly improved local-regional control and disease-free survival for adjuvant cisplatin-based chemoradiotherapy in comparison with radiotherapy alone for patients treated with primary surgical resection found to have high-risk pathologic features, including positive margins, ECE, two or more nodal metastases, level 4 or 5 nodal involvement (EORTC only), or perineural or vascular invasion (EORTC only). Combined analysis of the two trials demonstrated improved survival only among patients with positive margins or ECE when treated with adjuvant chemoradiotherapy versus radiotherapy alone. Notably, 10-year data from RTOG 9501 confirmed benefit for OPC and non-OPC patients.

Recently, several retrospective analyses of single-institutional case series have called into question whether or not ECE is a biomarker for high-risk of disease recurrence for HPV-positive OPC. In a University of Pittsburgh series of 214 oral cavity and 137 OPC patients, the frequency of ECE in nodal metastases was similar in the two groups. However, ECE was associated with disease-specific survival in oral cavity, but not OPC. Similarly, a case-series at Washington University did not observe the presence of ECE (present in 80% of patients with nodal metastases) to be associated with recurrence among 220 p16-positive OPC patients treated with primary surgical resection. At a median follow-up of 59 months, 5-year disease-specific survival for all patients in this series was 93%. Disease recurrence was observed in 22 patients (10%), with only four patients experiencing regional recurrence. Factors found to be associated with recurrence in univariate analysis included age, T3-4 versus T1-T2 stage, five or more metastatic nodes, and the presence of lymphovascular invasion. However, because of the small number of events, the independent predictors of disease recurrence could not be determined. Notably, in an earlier analysis, ECE was positively associated with several other adverse pathologic features, including number and level of involved lymph nodes, perineural invasion, angioinvasion, and lymphatic invasion. Thus, the independent effects of ECE on local-regional recurrence would be difficult to evaluate in a study with few events. Because of the high disease control rates among patients with HPV-positive OPC, these studies had very low power to detect prognostic differences among groups.

These retrospective analyses appropriately call into question the definition and use of “high-risk” features for selection of HPV-positive OPC patients for benefit from adjuvant chemoradiotherapy. The effect of ECE on disease progression clearly differs by HPV status, at least in terms of absolute magnitude. It must be noted, however, that all of the studies had very low power, and therefore estimated HRs are more important for inferences from the data than tests of significance (eg, P values). These suggest that the relative benefit from chemotherapy may be similar among HPV-positive and HPV-negative patients with ECE. In the first report of the Washington University series, ECE seemed associated with diminished disease-free survival (HR, 2.54; 95% CI, 0.88–7.34) and diminished overall survival (HR, 4.22; 95% CI, 1.17–12.1) in univariate analysis. Gross soft tissue extension also seemed associated with worse disease-free survival (HR, 2.1; 95% CI, 0.63–7.4) and benefit from chemoradiotherapy (HR, 2.78; 95% CI, 0.14–54.68). A case-series from Memorial Sloan Kettering Cancer Center observed a significantly worse survival among HPV-negative OPC patients with than without ECE (41.2 vs 76.1%; P = .002) together with nonsignificant differences for HPV-positive patients (74 vs 91.2%; P = .12). Additionally, these retrospective analyses are subject to bias by indication, and therefore it is difficult to evaluate the effect of chemoradiotherapy versus radiotherapy on disease control. A multicenter, randomized clinical trial comparing adjuvant chemoradiotherapy with radiotherapy for p16-positive OPC patients with ECE led by Washington University is ongoing. Until the results from that clinical trial become available, National Comprehensive Cancer Center guidelines recommend the use of adjuvant chemoradiotherapy for patients with ECE, regardless of tumor HPV status.